File size: 4,849 Bytes
ae850ff
39858fe
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ae850ff
 
39858fe
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
---
license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
- generated_from_trainer
datasets:
- fleurs
metrics:
- wer
model-index:
- name: wav2vec2-turkish-300m-7
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: fleurs
      type: fleurs
      config: tr_tr
      split: test
      args: tr_tr
    metrics:
    - name: Wer
      type: wer
      value: 0.16677037958929683
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# wav2vec2-turkish-300m-7

This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the fleurs dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3236
- Wer: 0.1668

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 4
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 0.1
- num_epochs: 35
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch   | Step  | Validation Loss | Wer    |
|:-------------:|:-------:|:-----:|:---------------:|:------:|
| 3.7291        | 0.6983  | 500   | 1.2114          | 0.8908 |
| 1.1707        | 1.3966  | 1000  | 0.3888          | 0.4555 |
| 0.5042        | 2.0950  | 1500  | 0.2879          | 0.3270 |
| 0.2623        | 2.7933  | 2000  | 0.2653          | 0.3265 |
| 0.2012        | 3.4916  | 2500  | 0.2405          | 0.2778 |
| 0.1817        | 4.1899  | 3000  | 0.2555          | 0.2704 |
| 0.1394        | 4.8883  | 3500  | 0.2452          | 0.2647 |
| 0.1112        | 5.5866  | 4000  | 0.2426          | 0.2458 |
| 0.1047        | 6.2849  | 4500  | 0.2520          | 0.2634 |
| 0.0916        | 6.9832  | 5000  | 0.2417          | 0.2443 |
| 0.0902        | 7.6816  | 5500  | 0.2627          | 0.2427 |
| 0.075         | 8.3799  | 6000  | 0.2551          | 0.2320 |
| 0.0716        | 9.0782  | 6500  | 0.2607          | 0.2221 |
| 0.0661        | 9.7765  | 7000  | 0.2504          | 0.2338 |
| 0.0634        | 10.4749 | 7500  | 0.2552          | 0.2229 |
| 0.0583        | 11.1732 | 8000  | 0.2637          | 0.2249 |
| 0.0537        | 11.8715 | 8500  | 0.2627          | 0.2122 |
| 0.0535        | 12.5698 | 9000  | 0.2654          | 0.2148 |
| 0.0521        | 13.2682 | 9500  | 0.2665          | 0.2123 |
| 0.0491        | 13.9665 | 10000 | 0.2814          | 0.2176 |
| 0.0466        | 14.6648 | 10500 | 0.2785          | 0.2138 |
| 0.0445        | 15.3631 | 11000 | 0.2856          | 0.2075 |
| 0.0415        | 16.0615 | 11500 | 0.2750          | 0.2076 |
| 0.0405        | 16.7598 | 12000 | 0.2743          | 0.2045 |
| 0.0368        | 17.4581 | 12500 | 0.2770          | 0.2013 |
| 0.0374        | 18.1564 | 13000 | 0.2961          | 0.2043 |
| 0.0374        | 18.8547 | 13500 | 0.2851          | 0.2028 |
| 0.0322        | 19.5531 | 14000 | 0.2955          | 0.1961 |
| 0.0317        | 20.2514 | 14500 | 0.3053          | 0.1998 |
| 0.0306        | 20.9497 | 15000 | 0.2988          | 0.1960 |
| 0.0328        | 21.6480 | 15500 | 0.2873          | 0.1949 |
| 0.0299        | 22.3464 | 16000 | 0.3030          | 0.1921 |
| 0.0272        | 23.0447 | 16500 | 0.2902          | 0.1866 |
| 0.0286        | 23.7430 | 17000 | 0.2962          | 0.1879 |
| 0.0288        | 24.4413 | 17500 | 0.3114          | 0.1871 |
| 0.0253        | 25.1397 | 18000 | 0.3203          | 0.1844 |
| 0.0262        | 25.8380 | 18500 | 0.2993          | 0.1861 |
| 0.0238        | 26.5363 | 19000 | 0.3108          | 0.1812 |
| 0.0228        | 27.2346 | 19500 | 0.3143          | 0.1759 |
| 0.0235        | 27.9330 | 20000 | 0.3077          | 0.1780 |
| 0.0227        | 28.6313 | 20500 | 0.3099          | 0.1739 |
| 0.0212        | 29.3296 | 21000 | 0.3144          | 0.1730 |
| 0.0212        | 30.0279 | 21500 | 0.3165          | 0.1726 |
| 0.0211        | 30.7263 | 22000 | 0.3178          | 0.1708 |
| 0.0192        | 31.4246 | 22500 | 0.3172          | 0.1682 |
| 0.0193        | 32.1229 | 23000 | 0.3188          | 0.1693 |
| 0.0195        | 32.8212 | 23500 | 0.3255          | 0.1661 |
| 0.0179        | 33.5196 | 24000 | 0.3248          | 0.1668 |
| 0.0166        | 34.2179 | 24500 | 0.3261          | 0.1668 |
| 0.018         | 34.9162 | 25000 | 0.3236          | 0.1668 |


### Framework versions

- Transformers 4.40.0
- Pytorch 2.2.2+cu121
- Datasets 2.17.1
- Tokenizers 0.19.1